def deploy(
self,
initial_instance_count,
instance_type,
candidate=None,
sagemaker_session=None,
name=None,
endpoint_name=None,
tags=None,
wait=True,
update_endpoint=False,
vpc_config=None,
enable_network_isolation=False,
model_kms_key=None,
):
"""Deploy a candidate to a SageMaker Inference Pipeline and return a Predictor
Args:
initial_instance_count (int): The initial number of instances to run
in the ``Endpoint`` created from this ``Model``.
instance_type (str): The EC2 instance type to deploy this Model to.
For example, 'ml.p2.xlarge'.
candidate (CandidateEstimator or dict): a CandidateEstimator used for deploying
to a SageMaker Inference Pipeline. If None, the best candidate will
be used. If the candidate input is a dict, a CandidateEstimator will be
created from it.
sagemaker_session (sagemaker.session.Session): A SageMaker Session
object, used for SageMaker interactions (default: None). If not
specified, one is created using the default AWS configuration
chain.
name (str): The pipeline model name. If None, a default model name will
be selected on each ``deploy``.
endpoint_name (str): The name of the endpoint to create (default:
None). If not specified, a unique endpoint name will be created.
tags (List[dict[str, str]]): The list of tags to attach to this
specific endpoint.
wait (bool): Whether the call should wait until the deployment of
model completes (default: True).
update_endpoint (bool): Flag to update the model in an existing
Amazon SageMaker endpoint. If True, this will deploy a new
EndpointConfig to an already existing endpoint and delete
resources corresponding to the previous EndpointConfig. If
False, a new endpoint will be created. Default: False
vpc_config (dict): Specifies a VPC that your training jobs and hosted models have
access to. Contents include "SecurityGroupIds" and "Subnets".
enable_network_isolation (bool): Isolates the training container. No inbound or
outbound network calls can be made, except for calls between peers within a
training cluster for distributed training. Default: False
model_kms_key (str): KMS key ARN used to encrypt the repacked
model archive file if the model is repacked
Returns:
callable[string, sagemaker.session.Session]: Invocation of
``self.predictor_cls`` on the created endpoint name.
"""
if candidate is None:
candidate_dict = self.best_candidate()
candidate = CandidateEstimator(candidate_dict,
sagemaker_session=sagemaker_session)
elif isinstance(candidate, dict):
candidate = CandidateEstimator(candidate,
sagemaker_session=sagemaker_session)
inference_containers = candidate.containers
endpoint_name = endpoint_name or self.current_job_name
return self._deploy_inference_pipeline(
inference_containers,
initial_instance_count=initial_instance_count,
instance_type=instance_type,
name=name,
sagemaker_session=sagemaker_session,
endpoint_name=endpoint_name,
tags=tags,
wait=wait,
update_endpoint=update_endpoint,
vpc_config=vpc_config,
enable_network_isolation=enable_network_isolation,
model_kms_key=model_kms_key,
)
def create_model(
self,
name,
sagemaker_session=None,
candidate=None,
vpc_config=None,
enable_network_isolation=False,
model_kms_key=None,
predictor_cls=None,
inference_response_keys=None,
):
"""Creates a model from a given candidate or the best candidate
from the automl job
Args:
name (str): The pipeline model name.
sagemaker_session (sagemaker.session.Session): A SageMaker Session
object, used for SageMaker interactions (default: None). If not
specified, the one originally associated with the ``AutoML`` instance is used.:
candidate (CandidateEstimator or dict): a CandidateEstimator used for deploying
to a SageMaker Inference Pipeline. If None, the best candidate will
be used. If the candidate input is a dict, a CandidateEstimator will be
created from it.
vpc_config (dict): Specifies a VPC that your training jobs and hosted models have
access to. Contents include "SecurityGroupIds" and "Subnets".
enable_network_isolation (bool): Isolates the training container. No inbound or
outbound network calls can be made, except for calls between peers within a
training cluster for distributed training. Default: False
model_kms_key (str): KMS key ARN used to encrypt the repacked
model archive file if the model is repacked
predictor_cls (callable[string, sagemaker.session.Session]): A
function to call to create a predictor (default: None). If
specified, ``deploy()`` returns the result of invoking this
function on the created endpoint name.
inference_response_keys (list): List of keys for response content. The order of the
keys will dictate the content order in the response.
Returns:
PipelineModel object
"""
sagemaker_session = sagemaker_session or self.sagemaker_session
if candidate is None:
candidate_dict = self.best_candidate()
candidate = CandidateEstimator(candidate_dict, sagemaker_session=sagemaker_session)
elif isinstance(candidate, dict):
candidate = CandidateEstimator(candidate, sagemaker_session=sagemaker_session)
inference_containers = candidate.containers
self.validate_and_update_inference_response(inference_containers, inference_response_keys)
# construct Model objects
models = []
for container in inference_containers:
image_uri = container["Image"]
model_data = container["ModelDataUrl"]
env = container["Environment"]
model = Model(
image_uri=image_uri,
model_data=model_data,
role=self.role,
env=env,
vpc_config=vpc_config,
sagemaker_session=sagemaker_session or self.sagemaker_session,
enable_network_isolation=enable_network_isolation,
model_kms_key=model_kms_key,
)
models.append(model)
pipeline = PipelineModel(
models=models,
role=self.role,
predictor_cls=predictor_cls,
name=name,
vpc_config=vpc_config,
sagemaker_session=sagemaker_session or self.sagemaker_session,
)
return pipeline